Art project translates music from Teenage Engineering's OP-Z synth into AI-generated imagery

The digital extension turns composed music into text prompts for Stable Diffusion.


AI-generated art is a new frontier rife with potential. But for every thorny question about copyright and the potential for widespread manipulation, generated art can also inspire wonder and awe. For example, look no further than this AI-powered experiment that creates kaleidoscopic visual landscapes for composed music.

A collaboration between quirky synth and hardware brand Teenage Engineering and design studios Modem and Bureau Cool, the project draws inspiration from the neurological condition synesthesia. This rare phenomenon leads the brain to perceive sensory input for several senses instead of one. For example, a listener with synesthesia may see music instead of only hearing it, observing color, movement and shape in response to musical patterns. Conversely, a synesthetic person may taste shapes, feel words from a novel or hear an abstract painting.

The audiovisual experiment uses the Teenage Engineering OP-Z sequencer as the music source that is then translated into AI art. In real-time, Modem and Bureau Cool’s “digital extension” translates musical properties into text prompts describing colors, shapes and movements. Those prompts then feed into Stable Diffusion (an open-source tool similar to DALL-E 2 and Midjourney) to produce dreamy and synesthetic animations.

Modem co-founder Bas van de Poel sees the experiment as fuel for artists’ imaginations. “With the project, we see the potential for musicians to explore new forms of creativity, facilitating a joint performance between human and machine,” van de Poel told Engadget.

If you’re a musician who owns Teenage Engineering’s OP-Z, you can’t yet use the extension yourself — but that may eventually change. Van de Poel tells Engadget that the companies are “exploring the potential of launching a public version.”

This AI-based project isn’t the first to bring synesthetic properties to the masses. Last year, Google Arts & Culture created an exhibition that flipped the concept around, bringing machine-learning-produced sound to Vassily Kandinsky’s paintings.